[bug] questions in Mertic:issue duration resolution

[longyanz]

Hi @frank-zsy ,when I read duration resolution metric in handbook.ipynb, I found this metric return a value about 4700 days. I think this maybe wrong

So I find some issues in our clickhouse and there issue open event missed. I think this caused the wrong result.

(microsoft/vscode#122346)
and in this clause, open_at field return 1970-1-1 .

[frank-zsy]

It's because the opened action was lost in the event log, so opened_at return as a null value which is 1970-1-1 when the type is DateTime. We should consider this since there may some lost data in the event log.

When the opened_at can not be found, we may use the first comment event instead, does this make sense to you?

[longyanz]

Thanks！

[frank-zsy]

It is really tricky if some events lost in the log, I will give a detail look into the problem later.

/self-assign

[xgdyp]

Hi @frank-zsy , why we don't use issue_created_at and issue_closed_at field to avoid this problem?

[frank-zsy]

Hi @frank-zsy , why we don't use issue_created_at and issue_closed_at field to avoid this problem?

Yes! Great point, I think I just forgot the uncommonly used columns which is quite useful in this situation.

[longyanz]

Hi, we use argMax(issue_created_at,created_at) AS opened_at instead of maxIf(created_at, action = 'opened') AS closed_at

It seems the result is correct.

[frank-zsy]

@longyanz Great, can you fix this?

[longyanz]

I will fix it after @xgdyp finish separating python kernel and nodejs kernel #1008 ~